unfunded liabilties' and uncertain fiscal financing · demographic shifts and rising relative...
TRANSCRIPT
“Unfunded Liabilities” and Uncertain Fiscal Financing Troy Davig, Eric M. Leeper, and Todd B. Walker
March 2010
RWP 10-09
“Unfunded Liabilities” and Uncertain Fiscal Financing
Troy Davig†, Eric M. Leeper‡, Todd B. Walker§
March 2010
RWP 10-09
Abstract A rational expectations framework is developed to study the consequences of alternative means to resolve the “unfunded liabilities” problem—unsustainable exponential growth in federal Social Security, Medicare, and Medicaid spending with no plan to finance it. Resolution requires specifying a probability distribution for how and when monetary and fiscal policies will change as the economy evolves through the 21st century. Beliefs based on that distribution determine the existence of and the nature of equilibrium. We consider policies that in expectation combine reaching a fiscal limit, some distorting taxation, modest inflation, and some reneging on the government’s promised transfers. In the equilibrium, inflation-targeting monetary policy cannot successfully anchor expected inflation. Expectational effects are always present, but need not have large impacts on inflation and interest rates in the short and medium runs. JEL Classification: H60, E30, E62, H30 Keywords: Fiscal sustainability, Inflation, Fiscal limit
Prepared for the Carnegie-Rochester Conference Series on Public Policy, “Fiscal Policy in an Era of Unprecedented Budget Deficits,” November 13-14, 2009. We thank our discussant, Kent Smetters, and the editor, Andy Abel, for suggestions that tightened the paper’s arguments considerably and we thank Hess Chung, Alex Richter, and Shu-Chun S. Yang for helpful conversations. The views expressed herein are those of the authors and do not necessarily represent those of the Federal Reserve Bank of Kansas City or the Federal Reserve System. † Federal Reserve Bank of Kansas City, [email protected] ‡ Corresponding author, Indiana University and NBER, 105 Wylie Hall, Bloomington, IN 47405; phone: (812) 855-9157; [email protected]; § Indiana University, walkerb@indiana.
Other Federal Non-interest Spending
Medicare and Medicaid
Social Security
1962 1972 1982 1992 2002 2012 2022 2032 2042 2052 2062 2072 2082
0
5
10
15
20
25
30
35
Per
centa
ge
ofG
DP
Revenues Model
Figure 1: Projected and actual federal expenditures decomposed into Medicare, Medicaid,and Social Security spending and other non-interest spending. The solid line representsactual and projected revenues under the Extended-Baseline scenario. The dashed line is thecalibration of the transfers process. See appendix A.1 for details. Source: CongressionalBudget Office (2009b).
1 Introduction
Profound uncertainty surrounds monetary and fiscal policy behavior in the United States.
Even in normal times, the multiple objectives that guide Federal Reserve decisions and the
absence of any mandates to guide federal tax and spending policies conspire to make it very
difficult for private agents to form expectations of monetary and fiscal policies.
In the wake of the financial crisis and recession of 2007-2009, monetary and fiscal poli-
cies have not been normal and, as long-term projections by the Congressional Budget Office
(CBO) make plain, in the absence of dramatic policy changes, policies are unlikely to re-
turn to normalcy for generations to come. Figure 1 reports actual and CBO projections of
federal transfers due to Social Security, Medicaid, and Medicare as a percentage of GDP.
Demographic shifts and rising relative costs of health care combine to grow these transfers
from about 10 percent of GDP today to about 25 percent in 70 years. One much-discussed
consequence of this growth is shown in figure 2, which plots actual and CBO projections of
federal government debt as a share of GDP from 1790 to 2083. Relative to the future, the
debt run-ups associated with the Civil War, World War I, World War II, the Reagan deficits,
“Unfunded Liabilities” and Uncertain Fiscal Financing
and the current fiscal stimulus are mere hiccups.
Of course, the CBO’s projections are accounting exercises, not economic forecasts. The
accounting embeds the assumption that current policies will remain in effect over the projec-
tion period. Because current policies have no provision for financing the projected transfers,
the accounting exercise converts those transfers into government borrowing, with no resulting
adjustments in prices or economic activity.
What can we learn from figure 2? Very little because it depicts a scenario that cannot
happen. In an economy populated by at least some forward-looking agents who participate
in financial markets, long before an explosive trajectory for debt is realized, bond prices
will plummet, sending long-term interest rates skyward. Why, even though the information
contained in figures 1 and 2 has been known for many years, do investors continue to acquire
U.S. government bonds bearing moderate yields?1 Evidently, financial market participants
do not believe that current policies will continue indefinitely. That long-term interest rates
are not rising steadily suggests that participants also seem not to believe that future policy
adjustments are likely to generate substantial inflation in the near term.
Spending projections like figure 1 are difficult to rationalize in optimizing models without
strong and implausible assumptions about information.2 There are several potential margins
along which to rationalize the projections without abandoning the assumption that economic
agents are well informed and forward looking. Either the expenditures will be financed
somehow—agents expect that some policies will adjust to raise sufficient surpluses—or the
implied “liabilities” in figure 2 are not liabilities—the government is expected to (partially)
renege on its promises through “entitlements reforms.” A third margin arises because the
bulk of U.S. government debt used to finance deficits is nominal: surprise price-level increases
1For example, the CBO has produced similar projections for well over a decade [O’Neill (1996), Congres-sional Budget Office (2002)] and economists like Auerbach and Kotlikoff had been discussing the “cominggenerational storm” long before Kotlikoff and Burns’s (2004) book by that title appeared [(Auerbach andKotilikoff, 1987, ch. 11), Auerbach, Gokhale, and Kotlikoff (1994, 1995)].
2Few public policy issues have received as much media attention, economic analysis, and political pon-tification as the periodic pronouncements of the bankruptcy of Social Security and the explosive growth inmedical spending. Given the intense attention these issues have received, equilibria in which private agentsare unaware of the problems arising from these transfers programs are wholly implausible.
2
“Unfunded Liabilities” and Uncertain Fiscal Financing
1790
1800
1810
1820
1830
1840
1850
1860
1870
1880
1890
1900
1910
1920
1930
1940
1950
1960
1970
1980
1990
2000
2010
2020
2030
2040
2050
2060
2070
2080
0
100
200
300
400
500
600
700
800
Per
centa
ge
ofG
DP
Alternative Fiscal Financing
Extended-Baseline Scenario
Figure 2: CBO’s projections of Debt-to-GDP ratio under Alternative Fiscal and Extended-Baseline Scenarios. See appendix A.1 for details.
can revalue outstanding debt to be in line with the expected present value of surpluses.
American fiscal policy institutions do nothing to inform economic agents about which
margins are likely to be exploited and when the policy adjustments will occur. Modeling the
economic consequences of the uncertainty that uninformative fiscal policy institutions create
is this paper’s primary contribution.
1.1 Modeling Policy Uncertainty This paper develops a rational expectations frame-
work to study an environment in which the government may be unwilling or unable to finance
its entitlements commitments entirely through direct tax collections. The core model is a
neoclassical growth economy with an infinitely lived representative household, capital ac-
cumulation, elastic labor supply, costly price adjustment, monetary policy, distorting taxes
levied against labor and capital income, and a process describing the evolution of promised
transfers from the government. By construction, the equilibrium is dynamically efficient
so it is not feasible for the government to permanently rollover debt. Distorting taxes im-
ply that there is a natural economic fiscal limit—the peak of the Laffer curve—to revenue
growth. Spending commitments, like those depicted in figure 1, that grow exponentially
cannot be financed entirely through direct taxes. How policies will adjust to be consistent
3
“Unfunded Liabilities” and Uncertain Fiscal Financing
with a rational expectations equilibrium is the crux of the model’s uncertainty.
“Unfunded liabilities” is a term that gets bandied about in discussions of the U.S. fiscal
position. Public finance economists and government agencies distinguish between prefunded
and unfunded programs, where unfunded entails future funding, as in pay-as-you-go financ-
ing. By this definition, unfunded programs can be either sustainable or unsustainable. But
the “unfunded liabilities” problem is a problem because of worries that future streams of
revenues will be insufficient to support the projected expenditures—in other words, that
current expenditure policies are not sustainable [Peter G. Peterson Foundation (2009)]. To
connect to this alternative usage of the term, we use “unfunded” to mean that if current
entitlements promises were to be honored, there would be insufficient surpluses to prevent an
explosive path of federal debt.
Any study of long-term fiscal issues, including accounting exercises, is highly speculative,
building in a great many tenuous assumptions about economic and policy developments
many decades into the future. The discipline of general equilibrium analysis and rational
expectations, however, requires still more speculation. For the exploding promised transfers
path in figure 1 to be consistent with a rational expectations equilibrium, we must take stands
on how policy may adjust in the future, the contingencies under which it may adjust, and how
much economic agents know about those adjustments. But to move beyond hand-wringing
about “unsustainable policies” and make progress on the macroeconomic consequences of
alternative policy adjustments, we are compelled to take some such stands. This paper
posits beliefs about policy adjustments that are consistent with equilibrium and it derives
how those beliefs affect the evolution of the macro economy. Our aim is to describe potential
policies and their consequences—not to prescribe particular policy solutions; neither do we
examine the welfare implications of potential policies.
Into the neoclassical model we build several layers of uncertainty. The promised transfers
process is stochastic and persistent, and initially follows a stationary process. At a random
date, the transfers process switches to the explosive process labeled “Model” in figure 1,
4
“Unfunded Liabilities” and Uncertain Fiscal Financing
which is treated as an absorbing state for promised transfers. Both before and after the
switch to the explosive transfers process, tax policy passively adjusts the distorting tax rate
to try to stabilize government debt, while monetary policy is active, obeying a Taylor rule
that aims to target the inflation rate.3 Over this period the promised transfers are funded
by direct tax revenues and borrowing from the public.
If these policies were to persist indefinitely, with transfers growing as a share of GDP,
the tax rate would eventually reach the peak of the Laffer curve.4 At this fiscal limit, tax
revenues have reached their maximum and, with no financing help from the active monetary
policy, government debt would take on the kind of explosive trajectory shown in figure 2.
Rational agents would expect this outcome and refuse to accumulate debt that cannot be
backed by future surpluses and seigniorage revenues. No equilibrium exists.
Although the peak of the Laffer curve is the economic fiscal limit, it is likely that in the
United States the political limit will be reached at much lower tax rates. As the New York
Times reports, “. . . Republicans refuse to talk about tax increases and Democrats refuse to
talk about cutting entitlement programs. . . ” [Sanger (2010)]. Ideally, the political fiscal
limit would be ground out endogenously as a political economy equilibrium, which requires
moving well outside our representative-agent environment.5 Instead, we model the vagaries
of the political system that grinds out the fiscal limit as an additional layer of uncertainty
that has an endogenous component: as tax rates rise, agents place higher probability on
hitting the limiting tax rate and inducing regime change, but the exact date of the change
3Applying Leeper’s (1991) definitions to a fixed-regime version of our model, “active” monetary policytargets inflation, while “passive” monetary policy weakly adjusts the nominal interest rate in response toinflation; “active” tax policy sets the tax rate independently of government debt and “passive” tax policychanges rates strongly enough when debt rises to stabilize the debt-GDP ratio; “active” transfers policymakes realized transfers equal promised transfers, while “passive” transfers policy allows realized transfersto be less than promised.
4A monetary economy may also have a Laffer curve for seigniorage revenues, as in Sargent and Wallace(1981). We do not allow for a distinct regime in which the printing press becomes an important revenuesource because, like direct taxation, seigniorage cannot permanently finance growing transfers.
5In addition, general equilibrium models of political economy tend to predict outcomes sharply at oddswith observed policy behavior; for example, sovereign debt defaults are predicted to occur either implausiblyoften or at implausibly low debt-GDP ratios on the order of 10 percent [Aguiar and Gopinath (2006), Arellano(2008), Mendoza and Yue (2008)].
5
“Unfunded Liabilities” and Uncertain Fiscal Financing
and the precise policies that will be adopted remain uncertain.
When tax rates reach their limit, they remain fixed and one of two possible regimes is
realized: either the government partially reneges on its promised transfers, in effect making
transfers policy passive, or monetary policy switches from being active to being passive,
adjusting the nominal interest rate only weakly in response to inflation. At the fiscal limit,
a random variable determines which policy becomes passive. Passive transfers policy is
equivalent to entitlements reform; “unfunded liabilities” are no longer liabilities because the
government chooses not to honor some fraction of its original promises. When monetary
policy becomes passive, although realized transfers equal promised transfers, a different kind
of reneging occurs as the price level jumps unexpectedly and the real value of outstanding
government debt drops, in effect funding the liabilities.
After the economy reaches the fiscal limit, it never returns to a regime in which tax
rates are adjusted to stabilize debt. Instead, the economy bounces randomly between the
passive transfers regime and the passive monetary policy regime. So long as the probability
of transiting to a passive transfers policy is positive, the economy can remain in the pas-
sive monetary policy regime indefinitely because expected transfers will remain below actual
transfers.
1.2 What the Model Delivers We solve for a rational expectations equilibrium in
which economic agents know the true probability distributions governing policy behavior
and they observe current and past realizations of policy decisions, including the regimes that
produce the policies. In contrast to much existing work on long-term fiscal policy, agents do
not have perfect foresight and must form expectations over all possible future policy regimes.
These expectations are what allow an equilibrium to exist, even in the presence of exploding
promised government transfers. Expectations of possible future policy regimes also play an
important role in determining the nature of equilibrium in the prevailing regime. Davig and
Leeper (2006, 2007) show that the expectations formation effects that arise in some future
regime can spill over into the current regime to produce equilibrium behavior that would not
6
“Unfunded Liabilities” and Uncertain Fiscal Financing
arise if only the current regime were possible.
Expectations about future adjustments in monetary and fiscal policies play a central
role in creating sustainable policies. Long-run expectations, which are driven by beliefs
about policy behavior beyond the fiscal limit, are anchored by the knowledge that policy
will fluctuate between a regime in which debt is revalued through jumps in the price level
and a regime in which the government may renege on its promised transfers. Because debt
is expected to be stabilized in the long run, the pre-fiscal limit expansion in debt can be
consistent with equilibrium.
Key findings include:
1. In an environment with exploding promised transfers, the conventional policy mix—
monetary policy targets inflation and tax policy targets debt—cannot anchor private
expectations on those policy targets. Before the economy has hit the fiscal limit, rising
government debt raises tax rates and the probability of hitting the limit. As that
probability rises, actual and expected inflation both increase. Failure of the usual
mix to work in the usual way stems from sustainability considerations that create the
expectation that policy regime will have to change in the future.
2. Expectational effects stemming from beliefs about future policy adjustments are ubiq-
uitous, but need not have large impacts on inflation and real and nominal interest
rates in the short to medium runs. Uncertainty about the fiscal limit and future policy
adjustments serves to smooth the expectational effects.
Because changes in policy regime are intrinsically non-linear, we solve the full non-linear
model numerically using the monotone map method. To ensure that the numerical solution
is consistent with a rational expectations equilibrium, we check that the solution satisfies
the transversality conditions associated with the underlying optimization problem. Before
turning to that complex model, we develop intuition for the nature of the numerical solution
by solving analytically a model that is considerably simpler.
7
“Unfunded Liabilities” and Uncertain Fiscal Financing
2 Contacts with the Literature
One of the key contributions of the paper is to examine how uncertainty about fiscal and
monetary policy can affect equilibrium outcomes. Sargent (2006) depicts policy uncertainty
by replacing the usual probability triple, (Ω,F ,P), with (?, ?, ?). He argues that the “prevail-
ing notions of equilibrium” (i.e., exogenous state-contingent rules, time-inconsistent Ramsey
problems, Markov-perfect equilibria, and recursive political equilibria) all assume agents
have a complete description of the underlying uncertainty, while Knightian uncertainty or
ambiguity about future policy might be a better description of reality. The agents in our
model are rational and condition on a well specified probability triple. However, our model
has several layers of uncertainty uncommon to the literature, which bridges the gap between
a typical rational expectations model and the ideas of Sargent (2006).
The importance of understanding how uncertainty about monetary and fiscal policy
shapes economic outcomes is obviously not a new concept. However, most of the early
work introducing fiscal policy into modern dynamic general equilibrium models was con-
ducted either in a non-stochastic environment or under the assumption that agents had
perfect foresight [Hall (1971), Abel and Blanchard (1983)]. Bizer and Judd (1989) and Dot-
sey (1990) are among the first papers to model fiscal policy uncertainty in a dynamic general
equilibrium framework. These papers emphasize the importance of uncertainty when ex-
amining welfare considerations, fiscal outcomes such as debt dynamics, and the behavior of
investment, consumption, prices and interest rates. Like Bizer and Judd (1989) and Dot-
sey (1990), our paper argues for a more realistic stochastic process with respect to policy
variables. We also find that adding important layers of uncertainty has profound effects on
equilibrium.
In a series of papers, Auerbach and Hassett (1992, 2007), Hassett and Metcalf (1999)
ask questions related to those posed here: What are the short-run and long-run economic
effects of random fiscal policy? How does policy stickiness—infrequent changes in fiscal
8
“Unfunded Liabilities” and Uncertain Fiscal Financing
policy—affect investment and consumption decisions? With uncertainty about factors af-
fecting fiscal variables, like demographics, what is the range of possible policy outcomes? The
baseline model used by Auerbach, Hassett and Metcalf to answer some of these questions is a
dynamic stochastic overlapping generations model with uncertain lifetimes and policy stick-
iness. Auerbach and Hassett (2007) find that introducing time-varying lifetime uncertainty
and policy stickiness changes the nature of the equilibrium and of optimal policy dramati-
cally. Our framework for addressing these questions is quite different. We re-interpret their
concept of stickiness by assuming that several policy instruments can switch randomly ac-
cording to a Markov chain. Instead of a one-time change in policy that is known, agents
face substantial uncertainty about when policies will adjust and which policies will adjust.
This allows for a broad range of potential policy outcomes, as advocated by Auerbach and
Hassett.
One dimension in which our model suffers is that it does not take into account the
important generational and distributional effects emphasized in Auerbach and Kotilikoff
(1987), Kotlikoff, Smetters, and Walliser (1998, 2007), Imrohoroglu, Imrohoroglu, and Joines
(1995), and Smetters and Walliser (2004). The canonical model used in these papers is an
overlapping generations model with each cohort living for 55 periods. This model permits
rich dynamics in demographics—population-age distributions, increasing longevity—intra-
generational heterogeneity, bequest motives, liquidity constraints, earnings uncertainty, and
so forth; this approach also allows for flexibility in modeling fiscal variables and alternative
policy scenarios. Due to the richness and complexity of the model, only perfect foresight
equilibria (or slight deviations from perfect foresight) are computed. A majority of the papers
cited above analyze the distributional consequences of Social Security reform.6 While we
cannot assess the distributional effects of alternative policies in our model, we substantially
increase the complexity of the policy uncertainty faced by individuals.
6Volume 50 of the Carnegie-Rochester Conference Series on Public Policy published in 1999 was devotedto the sustainability of Social Security. Many of the papers published in that volume are similar in spirit tothis paper. In particular, Butler (1999) emphasizes the importance of modeling the uncertainty about thetiming of policy changes.
9
“Unfunded Liabilities” and Uncertain Fiscal Financing
The type of uncertainty we examine is similar in some respects to the institutional un-
certainty described by North (1990, p. 83): “The major role of institutions in a society is to
reduce uncertainty by establishing a stable (but not necessarily efficient) structure to human
interaction. The overall stability of an institutional framework makes complex exchange
possible across both time and space.”
3 The Full Model
We employ a conventional neoclassical growth model with sticky prices, distorting income
taxation, and monetary policy.
3.1 Households An infinitely lived representative household chooses Ct, Nt,Mt, Bt, Kt
to maximize
Et
∞∑
i=0
βi
[C1−σt+i
1− σ− χ
N1+ηt+i
1 + η+ ν
(Mt+i/Pt+i)1−κ
1− κ
](1)
with 0 < β < 1, σ > 0, η > 0, κ > 0, χ > 0 and ν > 0. Ct =[∫
1
0ct (j)
θ−1
θ dj] θ
θ−1
is a
composite good supplied by a final-good producing firm that consists of a continuum of
individual goods ct(j), Nt denotes time spent working, and Mt/Pt are real money balances.
The household’s budget constraint is
Ct+Kt+Bt
Pt+Mt
Pt≤ (1− τt)
(Wt
PtNt +Rk
tKt−1
)+(1− δ)Kt−1+
Rt−1Bt−1
Pt+Mt−1
Pt+λtzt+
Dt
Pt,
where Kt−1 is the capital stock available to use in production at time t, Bt is one-period
nominal bond holdings, Mt is nominal money holdings, τt is the distorting tax rate, Rkt is the
real rental rate of capital, Rt−1 is the nominal return to bonds, zt are lump-sum transfers
promised by the government, λt is the fraction of promised transfers actually received by the
household and Dt is profits.
10
“Unfunded Liabilities” and Uncertain Fiscal Financing
3.2 Firms There are two types of firms: monopolistically competitive intermediate goods
producers who produce a continuum of differentiated goods and competitive final goods
producers.
3.2.1 Production of Intermediate Goods Intermediate goods producing firm j
has access to a Cobb-Douglas production function, yt(j) = kt−1(j)αnt(j)
1−α, where yt(j) is
output of intermediate firm j and kt−1(j) and nt(j) are the amounts of capital and labor
the firm rents and hires. The firm minimizes total cost, wtnt(j) +Rkt kt−1(j), subject to the
production technology.
3.2.2 Price Setting A final goods producing firm purchases intermediate inputs at
nominal prices Pt (j) and produces the final composite good using the following constant-
returns-to-scale technology, Yt =[∫
1
0yt (j)
θ−1
θ dj] θ
θ−1
, where θ > 1 is the elasticity of sub-
stitution between goods. Profit-maximization by the final goods producing firm yields a
demand for each intermediate good given by
yt (j) =
(Pt (j)
Pt
)−θ
Yt. (2)
Monopolistically competitive intermediate goods producing firm j chooses price Pt(j) to
maximize the expected present-value of profits
Et
∞∑
s=0
βs∆t+sDt+s (j)
Pt+s, (3)
where, because the household owns the firms, ∆t+s is the representative household’s stochas-
tic discount factor, Dt (j) are nominal profits of firm j, and Pt is the nominal aggregate price
level. Real profits are
Dt (j)
Pt=
(Pt (j)
Pt
)1−θ
Yt −Ψt(j)
(Pt (j)
Pt
)−θ
Yt −ϕ
2
(Pt (j)
π∗Pt−1 (j)− 1
)2
Yt, (4)
11
“Unfunded Liabilities” and Uncertain Fiscal Financing
where ϕ ≥ 0 parameterizes adjustment costs, Ψt(j) is real marginal cost, and we have used
the demand function in (2) to replace yt (j) in firm j’s profit function. Price adjustment is
subject to Rotemberg (1982) quadratic costs of adjustment, which arise whenever the newly
chosen price, Pt(j), implies that actual inflation for good j deviates from the steady state
inflation rate, π∗. The first-order condition for the firm’s pricing problem is
0 = (1− θ)∆t
(Pt (j)
Pt
)−θ (
YtPt
)+ θ∆tΨt
(Pt (j)
Pt
)−θ−1(
YtPt
)− (5)
ϕ∆t
(Pt (j)
π∗Pt−1 (j)− 1
)(Yt
π∗Pt−1 (j)
)+ βEt
[ϕ∆t+1
(Pt+1 (j)
π∗Pt (j)− 1
)(Pt+1 (j) Yt+1
π∗Pt (j)2
)].
In a symmetric equilibrium, every intermediate goods producing firm faces the same
marginal costs, Ψt, and aggregate demand, Yt, so the pricing decision is the same for all
firms, implying Pt (j) = Pt. Steady-state marginal costs are given by Ψ = (θ − 1)/θ, with
Ψ−1 = µ, where µ is the steady-state markup of price over marginal cost.
Note that in (4) the costs of adjusting prices subtracts from profits for firm j. In the
aggregate, costly price adjustment shows up in the aggregate resource constraint as
Ct +Kt − (1− δ)Kt−1 +Gt −ϕ
2
(Pt (j)
πPt−1 (j)− 1
)2
Yt = Yt (6)
3.3 Fiscal Policy, Monetary Policy, and the Fiscal Limit Fiscal policy finances
purchases, gt, and actual transfers, λtzt, with income tax revenues, money creation, and the
sale of one-period nominal bonds. The government’s flow budget constraint is
Bt
Pt+Mt
Pt+ τt
(Wt
PtNt +Rk
tKt−1
)= gt + λtzt +
Rt−1Bt−1
Pt+Mt−1
Pt. (7)
We now describe the various layers of uncertainty surrounding tax, transfers, and mon-
etary policies. Figure 3 describes schematically how the uncertainty unfolds. The fiscal
authority sets promised transfers exogenously according to a Markov chain that starts with
12
“Unfunded Liabilities” and Uncertain Fiscal Financing
AM/PF/AT
Stationary
Transfers
AM/PF/AT
Non-stationary
Transfers
pL,t
q = .5
PM/AF/AT
Regime 2
AM/AF/PT
Regime 1
1-p11 1-p22
Fiscal
limit 1-q = .5
pZ
Figure 3: The unfolding of uncertainty about policy.
a stationary process for transfers, while monetary policy actively targets inflation, and tax
policy raises tax rates passively when debt rises. The transfers process then switches, with
probability pZ , to a non-stationary process that is an absorbing state, mimicking the expo-
tentially growing line labeled “Model” in figure 1. The transfers process, which operates for
t ≥ 0, is
zt =
(1− ρz) z + ρzzt−1 + εt for Sz,t = 1
µzt−1 + εt for Sz,t = 2
(8)
where zt = Zt/Pt, |ρz| < 1, µ > 1, µβ < 1, εt ∼ N(0, σ2z). Szt follows a Markov chain,
where the regime with exploding promised transfers is an absorbing state and pz denotes
the probability of a switch to the non-stationary transfers regime. The expected number of
years until the switch from the stationary to nonstationary regime is (1− pz)−1. In figure 3
the switch from the stationary to the non-stationary transfers policy occurs with probability
pz and is marked by the move from the clear circle to the lightly shaded (blue) circle.
When transfers grow exponentially and are financed by new debt issuances, for some time
taxes can rise to support the expanding debt. At some point, however, for economic reasons—
reaching the peak of the Laffer curve—or political reasons—the electorate’s intolerance of
the distortions associated with high tax rates—tax policy will reach the fiscal limit. We
13
“Unfunded Liabilities” and Uncertain Fiscal Financing
model that limit as setting the tax rate to be a constant, τt = τmax for t ≥ T , where T is
the date at which the economy hits the fiscal limit. The setting of τmax is important for the
long-run performance of the economy and, a priori, there is no obvious value for τmax. One
possibility is that tax rates settle at a relatively high value; but another equally plausible
scenario is that taxpayers will revolt against high rates and τmax will be relatively low.
Tax policy sets rates according to
τt =
τ ∗ + γ(Bt−1/Pt−1
Yt−1
− b∗)
for Sτ,t = 1, t < T (Below Fiscal Limit)
τmax for Sτ,t = 2, t ≥ T (Fiscal Limit)
(9)
where b∗ is the target debt-output ratio and τ ∗ is the steady-state tax rate.
The date at which the fiscal limit is hit, T, is stochastic. We posit that the probability
at time t, pLt, of hitting the fiscal limit is an increasing function of the tax rate at time t−1,
implying that the probability rises with the level of government debt. A logistic function
governs the evolution of the probability
pL,t =exp(η0 + η1 (τt−1 − τ))
1 + exp(η0 + η1 (τt−1 − τ )), (10)
where η1 < 0. Households know the maximum tax rate, τmax, but the precise timing of when
that rate takes effect is uncertain.
Monetary policy behavior is conventional. It sets the short-term nominal interest rate in
response to deviations of inflation from its target
Rt = R∗ + α(πt − π∗) (11)
where π∗ is the target inflation rate. Monetary policy is active when α > 1/β, so policy
satisfies the Taylor principle. Policy is passive when 0 ≤ α < 1/β.
Once the economy reaches the fiscal limit, tax policy is active, as it is no longer feasible
14
“Unfunded Liabilities” and Uncertain Fiscal Financing
AM/AF/PT PM/AF/ATRegime 1 Regime 2
AM: Rt = R∗ + α(πt − π∗) PM: Rt = R∗
AF: τt = τmax AF: τt = τmax
PT: λtzt AT: zt
Table 1: AM: active monetary policy; PM: passive monetary policy; AF: active tax policy;AT: active transfers policy; PT: passive transfers policy.
to raise tax rates to finance further debt expansions. Because we take outright debt default
off the table, some other policy adjustments must occur. Given active tax policy, we restrict
attention to two possible policy regimes. One regime—denoted AM/AF/PT, Regime 1 in
table 1—has monetary policy actively targeting inflation, while transfers policy is passive,
with the government (partially) reneging on its promised transfers by the fraction λt ∈ [0, 1].
The other regime—denoted PM/AF/AT, Regime 2—has monetary policy abandoning the
targeting of inflation by pegging the nominal interest rate at R∗, and an “active” transfers
policy that sets actual transfers equal to promised transfers (λt = 1).
At the fiscal limit, the dark shaded (red) ball in figure 3, households put equal probability
on going to the two regimes in table 1, q = 1 − q = 0.5. Then policy bounces randomly
between those two regimes, labeled in the figure Regimes 1 and 2, governed by the transition
matrix
ΠT =
p11 1− p11
1− p22 p22
, (12)
where pii is the probability of remaining in regime i and 1− pii is the probability of moving
from regime i to regime j.
3.4 Calibration The parameters describing preferences, technology and price adjust-
ment are set to be consistent with Rotemberg and Woodford (1997) and Woodford (2003).
To study the impact of fiscal policy over a relatively long horizon, we calibrate the model
at an annual frequency. Intermediate-goods producing firms set the price of their good 15
percent over marginal cost, implying µ = θ(1 − θ)−1 = 1.15. For the price adjustment pa-
15
“Unfunded Liabilities” and Uncertain Fiscal Financing
rameter, we set ϕ = 10. If 66 percent of firms cannot reset their price each period, then a
calibration at a quarterly frequency would suggest ϕ would be around 70. Prices are cer-
tainly more flexible at an annual frequency, so the choice of ϕ = 10 captures a modest price
of cost adjustment. The annual real interest rate is set to 2 percent (β = .98). Preferences
over consumption and leisure are logarithmic, so σ = 1 and η = −1. χ is set so the steady
state share of time spent in employment is 0.33. Steady state inflation is set to 2 percent
and the initial steady state debt-output ratio is set to 0.4.
For real balances, δ is set so velocity in the deterministic steady state, defined as cP/M,
matches average U.S. monetary base velocity at 2.4. We take this value from Davig and
Leeper (2006), based on data from 1959-2004 on the average real level of expenditures on
non-durable consumption plus services. The parameter governing the interest elasticity of
real money balances, κ, is set to 2.6 [Chari, Kehoe, and McGrattan (2000)].
Mean federal government purchases are set to a constant 8 percent share of output. In
the stationary transfer regime, z is set to make steady state transfers 9 percent of output
and the process for transfers is persistent, with ρZ = .9. Monetary policy is active in the
stationary regime, where α = 1.5, and fiscal policy is passive, γ = .15. The inflation target
is π∗ = 2.0. Steady state debt to output in this regime is set to .4, which determines b∗.
This implies that τ = .1980, a value for the average tax rate in line with figure 1. The
expected duration of the stationary regime is five years (pz = .8), which is roughly the time
before the CBO projects that transfers begin their upward trajectory. Transfers then grow
at 1 percent per year once the switch from the stationary to non-stationary regime occurs,
making µ = 1.01.
Once the economy switches to the regime with rising transfers, the same monetary and
fiscal rules stay in place until the economy hits the fiscal limit. The probability of hitting
the fiscal limit rises according to the logistic function (10), where η0 and η1 are set so that
initially, the probability of hitting the fiscal limit is 2 percent. The probability then rises as
debt and taxes rise, reaching roughly 20 percent by 2075, after which it rises rapidly.
16
“Unfunded Liabilities” and Uncertain Fiscal Financing
At the fiscal limit, we assume tax rates are constant, τmax = .2425. In the original
stationary transfer state, this tax rate would support a steady state debt-output ratio of 2.3,
an unprecedented level for the United States. However, with rising transfers, the level at
which debt stabilizes is well below this. At the fiscal limit, higher tax rates are no longer
available to stabilize debt, so we allow two potential resolutions, each with a 50 percent
chance of being realized. The first resolution is a switch to passive monetary policy, where
the monetary authority pegs the nominal interest rate (α = 0) and the fiscal authority
continues to make good on its promised transfers. The second resolution involves reneging
on promised transfers payments. Under this scenario, monetary policy remains active. Each
regime is calibrated to be persistent, where the expected duration of the passive monetary
regime is 10 years and 100 years for the reneging regime.
4 Some Analytical Intuition
Before launching into the numerical solution of the complicated model in section 3, we
highlight some of the mechanisms at work in the numerical solution with a simpler setup that
admits an analytical solution. The setup is stark. It eliminates virtually all the uncertainty
about future policies by assuming that the government’s promised transfers always follow a
non-stationary process and by positing that at some known future date, T , fiscal policy hits
the fiscal limit and switches from passively adjusting lump-sum taxes to stabilize debt, to
fixing tax revenues at their limit, τmax. At date T , one of two possible policy combinations is
pursued: either the government reneges on promised transfers or monetary policy becomes
passive and pegs the nominal interest rate. Agents know which policy combination will be
realized at T.
Consider a constant endowment economy that is at the cashless limit. A representative
household pays lump-sum taxes, τt, receives lump-sum transfers, zt, and chooses sequences
of consumption and risk-free nominal bonds paying gross nominal interest Rt, ct, Bt, to
maximize E0
∑∞
t=0βtu(ct) subject to the budget constraint ct + Bt/Pt + τt = yt + zt +
17
“Unfunded Liabilities” and Uncertain Fiscal Financing
Rt−1Bt−1/Pt, with R−1B−1 > 0 given. Government purchases are zero in each period so that
goods market clearing implies ct = yt = y. In equilibrium the consumption Euler equation
reduces to the simple Fisher relation
1
Rt= βEt
(PtPt+1
)(13)
where β ∈ (0, 1) is the household’s discount factor. With a constant endowment, the equi-
librium real interest rate is also constant at 1/β.
We imagine that the government’s promised transfers follow the non-stationary process
given by (8), zt = µzt−1 + εt, where µ > 1, Etεt+1 = 0, and z−1 > 0 is given. The process for
promised transfers holds for all t ≥ 0. The growth rate of transfers is permitted to be positive,
but it must be bounded to ensure that µβ < 1. This restriction implies that transfers cannot
grow faster than the steady state real interest rate.
Explosive growth in transfers implies that at some date T in the future, the economy
will reach the fiscal limit, beyond which households will be unwilling to hold additional debt
and the government has reached its maximum capacity to levy taxes. This is not strictly
true in this endowment economy with lump-sum taxes, but it would be true in a production
economy with distorting taxation, such as the model in section 3.7 Date T is the period at
which the tax rate reaches its fiscal limit, so that τt = τmax for t ≥ T , as in (9).
In the period before the economy reaches the fiscal limit, monetary policy is active and
fiscal policy is passive. The rules are similar to those described in section 3,
R−1
t = R∗−1 + α
(Pt−1
Pt−
1
π∗
), α > 1/β (14)
for monetary policy, where π∗ is the inflation target. Fiscal policy adjusts taxes passively to
7In this model with lump-sum taxes there is no upper bound for taxes or debt, so long as debt does notgrow faster than the real interest rate. But in a more plausible production economy, in which taxes distortbehavior, there would be a natural fiscal limit. See Bi (2009) for an application of an endogenous fiscal limitto the issue of sovereign debt default.
18
“Unfunded Liabilities” and Uncertain Fiscal Financing
the state of government debt
τt = τ ∗ + γ
(Bt−1
Pt−1
− b∗), γ > r = 1/β − 1 (15)
where b∗ is the debt target.
The government’s flow budget constraint is
Bt
Pt+ τt = zt +
Rt−1Bt−1
Pt(16)
If policy rules (14) and (15) were to remain in effect forever, the promised transfers pro-
cess were stationary, and there were no fiscal limit, this economy would exhibit Ricardian
equivalence, with inflation determined only by parameters describing monetary policy be-
havior. But, with a non-stationary transfers process and a fiscal limit, such policy behavior,
if agents believed it would remain in effect permanently, would produce an explosive path
for government debt that, in finite time, would hit the fiscal limit and no equilibrium would
exist. A rational expectations equilibrium requires that some policies adjust.
At the fiscal limit, τt = τmax, t ≥ T, and monetary and transfers policies adjust. We
examine two limiting cases. In the first, monetary policy continues to obey (14), but the
government does not fully honor its promised transfers; instead, it delivers λtzt, for t ≥ T . In
the second case, transfer promises are honored, λ = 1, t ≥ T, but monetary policy abandons
the Taylor principle to simply peg the nominal rate at Rt = R∗, t ≥ T .
4.1 A Reneging Regime Allowing the government to renege on promised transfers
completely neutralizes the fiscal limit. If λt is permitted to be negative, then transfers can
exactly mimic the fiscal financing activities of lump-sum taxes and the economy exhibits
Ricardian equivalence.
Modify the government budget constraint for periods t ≥ T to be Bt
Pt+ τmax = λtzt +
Rt−1Bt−1
Pt, and set λt = −γbt−1/zt with γ > r, as in (15). Equilibrium inflation comes from
19
“Unfunded Liabilities” and Uncertain Fiscal Financing
combining (13) and (14) when monetary policy is active to yield
β
αEt
(PtPt+1
−1
π∗
)=Pt−1
Pt−
1
π∗
(17)
Aggressive reactions of monetary policy to inflation imply that β/α < 1 and the unique
bounded solution for inflation is πt = π∗, so equilibrium inflation is always on target, as is
expected inflation.
If monetary policy determines inflation and the path of the price level, Pt, how must
fiscal policy respond to disturbances in transfers to ensure that policy is sustainable? This
is where passive transfers adjustments step in. Substituting the tax rule, (15), into the
government’s budget constraint, taking expectations conditional on information at t − 1,
and employing the Fisher relation, (13), yields the expected evolution of real debt
Et−1
(Bt
Pt
)+ τmax = Et−1λtzt + (β−1 − γ)
(Bt−1
Pt−1
)(18)
Because β−1 − γ < 1, higher debt brings forth the expectation of lower transfers, so (18)
describes how debt is expected to return to steady state following a shock to zt.
4.2 A Regime Where Monetary Policy Cannot Control Inflation Iterating
forward on the government budget constraint and imposing both the household’s transver-
sality condition for debt accumulation and the household’s Euler equation, we obtain the
intertemporal equilibrium condition
B0
P0
= E0
∞∑
j=1
βjsj (19)
where st ≡ τt − zt is the primary surplus.
To solve for this equilibrium we break the intertemporal equilibrium condition into two
20
“Unfunded Liabilities” and Uncertain Fiscal Financing
parts
B0
P0
= E0
T−1∑
j=1
βjsj + E0
∞∑
j=T
βjsj (20)
where the function for the primary surplus, st, changes at the fiscal limit
st =
τ ∗ − γ(Bt−1/Pt−1 − b∗)− zt, t = 0, 1, ..., T − 1
τmax − zt, t = T, ...,∞
(21)
Expression (20) decomposes the value of government debt at the initial date into the expected
present value of surpluses leading up to the fiscal limit and the expected present value of
surpluses after the limit has been hit.
Evaluating the second part of (20) and letting τmax = τ ∗, after the limit is hit at T
E0
∞∑
j=T
βjsj = E0
(BT−1
PT−1
)=
βT
1− β(τ ∗ − z∗)−
(βµ)T
1− βµ(z0 − z∗) (22)
The first part of (20) is a bit messier, as it involves solving out for the endogenous taxes
that are responding to the state of government debt before the fiscal limit is hit. That part
of (20) may be written as
E0
T−1∑
j=1
βjsj = (τ ∗ − γb∗ − z∗)T−1∑
j=1
(β
1− γβ
)j
− (z0 − z∗)T−1∑
j=1
(βµ
1− γβ
)j
(23)
Pulling together (22) and (23) yields equilibrium real debt at date t = 0 as a function of
fiscal parameters and the date 0 realization of transfers
B0
P0
=
[βT
1− β+
T−1∑
j=1
(β
1− γβ
)j ](τ ∗ − z∗)− γb∗
T−1∑
j=1
(β
1− γβ
)j
−
[(βµ)T
1− βµ+
T−1∑
j=1
(βµ
1− γβ
)j ](z0 − z∗) (24)
21
“Unfunded Liabilities” and Uncertain Fiscal Financing
This expression determines the equilibrium value of debt. The value of debt at t = 0—and
by extension at each date in the future—is uniquely determined by parameters describing
preferences and fiscal behavior and by the exogenous realization of transfers at that date.
We arrive at this expression by substituting out the endogenous sequence of taxes before
the fiscal limit. Apparently this economy will not exhibit Ricardian equivalence even if tax
policy obeys a rule that raises taxes to retire debt back to steady state. This occurs despite
the fact that in the absence of a fiscal limit such a tax rule delivers Ricardian equivalence.
Two other implications are immediate. Higher transfers at time 0, z0, which portend a
higher future path of transfers because of their positive serial correlation, reduce the value
of debt. This occurs because higher expected government expenditures reduce the backing
and, therefore, the value of government liabilities. A second immediate implication is more
surprising. How aggressively tax policy responds to debt before hitting the fiscal limit, as
parameterized by γ, matters for the value of debt. Permanent active monetary/passive tax
policies, in contrast, produce Ricardian equivalence in this model, so the timing of taxation
is irrelevant: how rapidly taxes stabilize debt has no bearing on the value of debt. Both
of these unusual implications are manifestations of the breakdown in Ricardian equivalence
that occurs when there is the prospect that at some point the economy will hit a fiscal limit,
beyond which taxes will no longer adjust to finance debt.8
We now turn to how the equilibrium price level is determined. Given B0/P0 from (24)
and calling the right side of (24) b0, use the government’s flow budget constraint at t = 0
and the fact that s0 = τ0 − z0 to solve for P0:
P0 =R−1B−1
b0 + τ0 − z0(25)
Given R−1B−1 > 0, (25) yields a unique P0 > 0. Entire sequences of equilibrium Pt, R−1t ∞t=0
8This is not to suggest that one cannot concoct “Ricardian scenarios.” For example, because T is known,if the government were to commit to fully finance the change in the present value of transfers that arisesfrom a shock to z0 before the economy reaches the fiscal limit, one could obtain a Ricardian outcome. Butthis is driven by the fact that T is known. If T were uncertain, as in the model in section 3, with someprobability of occurring at every date, even this cooked-up scenario would not produce a Ricardian result.
22
“Unfunded Liabilities” and Uncertain Fiscal Financing
are solved recursively: having solved for B0/P0 and P0, obtain R0 from the monetary policy
rule and derive the nomimal value of debt. Then use (24) redated at t = 1 to obtain
equilibrium B1/P1 and the government budget constraint at t = 1 to solve for P1 using (25)
redated at t = 1, and so forth.
The equilibrium price level has the same features as it does under a fiscal theory equi-
librium [Leeper (1991), Sims (1994), Woodford (1995), Cochrane (1999)]. Higher current
or expected transfers, which are not backed in present-value terms by expected taxes, raise
household wealth, increase demand for goods, and drive up the price level (reducing the value
of debt). A higher pegged nominal interest rate raises nominal interest payments, raising
wealth and the price level next period. Similarities between this equilibrium and those in
the fiscal theory literature stem from the fact that price-level determination is driven by
beliefs about policy in the long run. From T on, this economy is identical to a fixed-regime
passive monetary/active fiscal policies economy and it is beliefs about long-run policies that
determine the price level. Of course, before the fiscal limit the two economies are quite
different and the behavior of the price level will also be different.
In this environment where the equilibrium price level is determined entirely by fiscal
considerations through its interest rate policy, monetary policy determines the expected
inflation rate. Combining (13) with (14) we obtain an expression in expected inflation in
periods t < T
Et
(PtPt+1
−1
π∗
)=α
β
(Pt−1
Pt−
1
π∗
)(26)
where α/β > 1.
Given equilibrium P0 from (25) and an arbitrary P−1—arbitrary because the economy
starts at t = 0 and cannot possibly determine P−1, regardless of policy behavior—(26) shows
that E0(P0/P1) grows relative to the initial inflation rate. In fact, throughout the active
monetary policy/passive fiscal policy phase, for t = 0, 1, . . . , T − 1, expected inflation grows
at the rate αβ−1 > 1. In periods t ≥ T monetary policy pegs the nominal interest rate at
R∗, and expected inflation is constant: Et(Pt/Pt+1) = (R∗β)−1 = 1/π∗.
23
“Unfunded Liabilities” and Uncertain Fiscal Financing
The implications of the equilibrium laid out in equations (24), (25), and (26) for govern-
ment debt, inflation, and the anchoring of expectations on the target values (b∗, π∗) are most
clearly seen in a simulation of the equilibrium. Figure 4 contrasts the paths of the debt-
GDP ratio from two models: a fixed passive monetary/active tax regime—dashed line—and
the present model in which an active monetary/passive tax regime is in place until the
economy hits the fiscal limit at date T , when policies switch permanently to a passive mon-
etary/active tax combination—solid line.9 The fixed regime displays stable fluctuations of
real debt around the 50 percent steady state debt-GDP, which the other model also produces
once it hits the fiscal limit. Leading up to the fiscal limit, however, it is clear that the active
monetary/passive tax policy combination does not keep debt near target.
Expected inflation evolves according to (26). Leading up to the fiscal limit monetary
policy is active, with α > 1/β, and there is no tendency for expected inflation to be anchored
on the inflation target. Figure 5 plots the inflation rate from the same fixed-regime model—
dashed line—and from the present model—solid line—along with expected inflation from
the present model—dotted dashed line. Inflation in the fixed regime fluctuates around π∗
and, with the pegged nominal interest rate, expected inflation is anchored on target. But
in the period leading up to the fiscal limit, the price level is determined by the real value
of debt, which as figure 4 shows, deviates wildly from its target, b∗. Expected inflation in
that period, though not independent of the inflation target, is certainly not anchored by
the target. Instead, under active monetary policy, the deviation of expected inflation from
target grows with the deviation of actual inflation from target in the previous period. The
figure shows how equation (26) makes expected inflation follow actual inflation, with active
monetary policy amplifying movements in expected inflation.
Figures 4 and 5 also make the point that as the economy approaches the fiscal limit at
date T , the equilibrium converges to the fixed-regime economy. This result is driven by the
9Figures 4 and 5 use the following calibration. Leading up to the fiscal limit, α = 1.50 and γ = 0.10 andat the limit and in the fixed-regime model, α = γ = 0.0. We assume steady state values τ∗ = 0.19, z∗ = 0.17,π∗ = 1.02 (gross inflation rate) and we assume 1/β = 1.04 so that b∗ = 0.50. The transfers process hasρ = 0.90 and σ = 0.002. Identical realizations of transfers were used in all the figures.
24
“Unfunded Liabilities” and Uncertain Fiscal Financing
0 5 10 15 20 25 30 35 40 45 500.44
0.46
0.48
0.5
0.52
0.54
0.56
0.58
0.6
0.62
0.64
Debt−GDP Target
Fiscal LimitT = 50
Debt WhenFiscal Limit at
T = 50
Debt inRegime 2
Figure 4: Debt-GDP ratios for a particular realization of transfers for two models: a fixedpassive monetary/active tax regime—dashed line—and the present model in which an activemonetary/passive tax regime is in place until the economy hits the fiscal limit at date T ,when policy switch permanently to a passive monetary/active tax combination—solid line.
fact that date T is known.
To underscore the extent to which inflation is unhinged from monetary policy, even in
the active monetary/passive tax regime before the fiscal limit, suppose that tax policy reacts
more aggressively to debt. A higher value of γ in (15) can have unexpected consequences.
Expression (24) makes clear that raising γ, which in a fixed active monetary/passive tax
regime would stabilize debt more quickly, amplifies the effects of transfers shocks on debt.
A more volatile value of debt, in turn, translates into more volatile actual and expected
inflation.
An analogous exercise for monetary policy illustrates its impotence when there is a fiscal
limit. A more hawkish monetary policy stance, higher α in (14), has no effect whatsoever
on the value of debt and inflation: α does not appear in expression (24) for real debt or
expression (25) for the price level. More hawkish monetary policy does, however, amplify
the volatility of expected inflation.
Because monetary policy loses control of inflation after the fiscal limit it reached, forward-
looking behavior implies it also loses control of inflation before the fiscal limit is hit. By
25
“Unfunded Liabilities” and Uncertain Fiscal Financing
0 5 10 15 20 25 30 35 40 45 50
0.8
0.9
1
1.1
1.2
1.3
Expected InflationWhen Fiscal Limit
at T = 50 Inflation WhenFiscal Limitat T = 50
InflationTarget
Inflation inRegime 2
Fiscal LimitT = 50
Figure 5: Inflation for a particular realization of transfers for two models: a fixed pas-sive monetary/active tax regime—dashed line—and the present model in which an activemonetary/passive tax regime is in place until the economy hits the fiscal limit at date T ,when policy switch permanently to a passive monetary/active tax combination—solid line;expectation of inflation from present model—dotted dashed line.
extension, changes in fiscal behavior in the period leading up to the limit affect both the
equilibrium inflation process and the process for expected inflation.
5 Numerical Solution and Results
We solve the model laid out and calibrated in section 3 using a nonlinear algorithm called
the monotone path method described in Davig and Leeper (2006). That method discretizes
the state space and finds a fixed point in decision-rules for each point in the state space.
Details appear in appendix A.2.
This section reports results from the numerical solution. First, we show the transitional
dynamics of the model for a particular realization of monetary-tax-transfers regimes. These
dynamics highlight the critical role that expectations of future policies play in determining
equilibrium. Then we perform a Monte Carlo exercise, taking draws from policy regimes
according to the Markov processes described in section 3. This exercise yields the stationary
26
“Unfunded Liabilities” and Uncertain Fiscal Financing
distribution of the economy.
5.1 Transition Paths It is useful to condition on a particular realization of policy
regimes and trace out how the equilibrium evolves following each type of regime change.
Although in this counterfactual exercise we treat regimes as absorbing states, in the equilib-
ria we study, agents form expectations based on the true probability distributions for policy
regimes, as described in sections 3.3 and 3.4, and their decision rules embed those expecta-
tions. Expectations about policy regimes in the future can have strong effects on behavior
in the prevailing regime. Because agents form expectations rationally, the counterfactual
allows us to back out the expectational effects, which are reflected in sequences of forecast
errors—policy surprises—that generate the dynamic adjustments along the transition paths.
Realized policy regimes produce “shocks” to which agents in the model respond.10 One
cautionary note: because these counterfactuals impose regime changes at arbitrary dates,
rather than at dates governed by the underlying Markov processes, the counterfactuals are
most useful for understanding qualitative features of the equilibrium. Quantitative features
are discussed in the Monte Carlo exercise in section 5.2.
5.1.1 Active Monetary, Passive Tax, and Active Transfers Policies We imag-
ine that the economy starts in 2010 in a regime characterized by active monetary policy,
passive tax policy, and active (stationary) transfers policy and then switches in 2015 to an
identical regime, but where transfers follow the non-stationary process, Sz,t = 2 in expression
(8). When transfers policy is active, the government makes good on its promised path of
transfer payments. Transition paths shown in figure 6 condition on staying in the latter
regime throughout the simulation.
Rising debt and the resulting rising tax rates are the dominant forces in this scenario.
Steadily rising current and expected tax rates shift labor supply in, reducing hours worked,
10These are not identical to impulse response functions used to study linear models. In linear models,impulse response functions show the response of the economy to a one-time i.i.d. shock, with the pathfollowing the shocks being deterministic. In contrast, our counterfactuals create a sequence of shocks arisingfrom what agents perceive to be surprise realizations of policy regimes.
27
“Unfunded Liabilities” and Uncertain Fiscal Financing
2010 2020 2030 2040 2050 2060−4
−2
0
Output (% deviation from 2009)
2010 2020 2030 2040 2050 2060−10
−5
0
Consumption (% deviation from 2009)
2010 2020 2030 2040 2050 2060−5
0
5
10Capital Stock (% deviation from 2009)
2010 2020 2030 2040 2050 2060−8−6−4−2
0
Labor (% deviation from 2009)
2010 2020 2030 2040 2050 20600.2
0.22
0.24
0.26
0.28Tax Rate
2010 2020 2030 2040 2050 20600.5
1
1.5Debt / Output
2010 2020 2030 2040 2050 20601
2
3Inflation (%)
2010 2020 2030 2040 2050 20603
4
5
6Nominal Interest Rate (%)
τmax
Figure 6: Transition paths conditional on remaining in the regime with active monetary,passive tax, and active transfers policies so the economy does not hit the fiscal limit.
and discourage investment, reducing the capital stock below its 2009 steady-state level. Both
consumption and output decline. As the tax rate rises, the probability of hitting the fiscal
limit and switching to other policies also rises, according to the logistic function in (10).
The likelihood of moving to other regimes creates two expectational effects. First, while
the date at which the fiscal limit kicks in is uncertain, agents know that at the limit the tax
rate will be permanently fixed at τt = τmax, which in the simulation is 0.2425. When the
current rate exceeds τmax, agents expect lower rates in the future. Lower expected rates raise
the expected return to capital, raising capital accumulation at the expense of consumption.
This effect, however, does not kick in until about 2045 or later. Along the entire transition
path tax policy is forced to remain passive, although agents put positive probability on
hitting the limit. Since tax rates rise along with debt, labor also declines persistently.
The second expectational effect begins much sooner. Inflation rises in either of the
two future regimes to which the economy moves after the fiscal limit. This creates higher
expected inflation which has effects almost immediately, although they are small. Inflation
28
“Unfunded Liabilities” and Uncertain Fiscal Financing
in the model has two sources: higher expected inflation and higher real marginal costs.
Marginal cost in the model depends on the ratio of each factor price to its marginal product.
Over the horizon that inflation is rising in figure 6, real marginal cost is falling very gradually,
which would tend to lower inflation, implying that expected inflation drives the steady rise
in actual inflation. Although some of the increase in inflation is expected, over the period
when inflation rises, agents are continually surprised by inflation.
This expected inflation effect is a more sophisticated version of the phenomenon that
section 4 highlights. In the analytical example, the date of the fiscal limit is known, bringing
the effects of expected future policies strongly into the present. In figure 6 the steady rise
in inflation is driven by the steady rise in the probability of hitting the fiscal limit and
switching to one of the other two regimes. A time-varying probability of reaching the fiscal
limit translates directly into time-varying expected inflation.
Despite the sequence of positive inflation surprises before 2045, nominal government
debt does not get revalued and stabilized. Active monetary policy prevents revaluation by
adjusting the nominal interest rate more than one-for-one with inflation. In periods of rising
(surprise) inflation, monetary policy effectively raises nominal debt service by more than the
inflation rate, ensuring continued growth in debt.
The upturn in the capital stock that begins in 2045 is also driven by two distinct ex-
pectational effects. As mentioned, one is the expectation of lower tax rates in the future.
Another expectational effect derives from the possibility of switching to the regime in which
the government does not fully honor its promised transfers. Lower expected transfers encour-
age savings in the form of capital accumulation. A higher capital stock reduces its marginal
product sufficiently that real marginal costs begin to decline. Although expected inflation
continues to rise, the falling marginal costs eventually dominate, bringing inflation down
quickly.
From figure 6 emerges a central message of the paper: in an environment in which fiscal
policy is unwilling or unable to stabilize debt, monetary policy cannot successfully target
29
“Unfunded Liabilities” and Uncertain Fiscal Financing
2046 2048 2050 2052 2054 2056 2058−6
−4
−2
0
Output (% deviation from 2009)
2046 2048 2050 2052 2054 2056 2058−10
−5
0
Consumption (% deviation from 2009)
2046 2048 2050 2052 2054 2056 2058−10
0
10
Capital Stock (% deviation from 2009)
2046 2048 2050 2052 2054 2056 2058−6
−4
−2
0
2
Labor (% deviation from 2009)
2046 2048 2050 2052 2054 2056 2058
0.24
0.26
Tax Rate
2046 2048 2050 2052 2054 2056 2058
0.5
1
1.5Debt / Output
2046 2048 2050 2052 2054 2056 20580
2
4
Inflation (%)
2046 2048 2050 2052 2054 2056 20583
4
5
6Nominal Interest Rate (%)
τmax
Figure 7: Transition paths for other two regimes superimposed on paths conditional on ac-tive monetary, passive tax, and active transfers policies. Active monetary/passive tax/activetransfers policies are solid lines; active monetary/active tax/passive transfers policies out-comes are marked by triangles; passive monetary/active tax/active transfers policies out-comes are marked by squares. Large spikes in variables in the passive monetary/activetax/active transfers regime are truncated.
inflation. The active monetary policy regime that prevails along the transition path responds
aggressively to inflation with the aim of hitting a 2 percent inflation target. For over four
decades it misses the target on the high side and then it misses the target on the low
side. Both misses are due to expectational effects that are beyond control of the monetary
authority in the current regime. Monetary policy loses control of inflation in an extreme form:
it cannot control either actual or expected inflation, even though it is hawkishly targeting
inflation.
5.1.2 Active Monetary, Active Tax, and Passive Transfers Policies This sce-
nario and the next arise after the economy hits the fiscal limit—year 2047 in the simulation—
and tax rates are permanently set at τmax. Unlike in the analytical model, in the numerical
30
“Unfunded Liabilities” and Uncertain Fiscal Financing
2036 2038 2040 2042 2044 2046 2048 2050 2052 2054 2056 205865
70
75
80
85
90
95
100
Figure 8: Realized path of λt, the percentage of promised transfers that is honored, associatedwith the counterfactual exercise in figure 7, which is conditional on active monetary, activetax, and passive transfers policies.
solution the economy does not then enter an absorbing state, as the two ultimate regimes
recur randomly.
Leading up to 2047 the path of the economy is identical to that discussed in section 5.1.1.
From 2047 on, tax policy is active—unresponsive to debt—and in the current scenario mon-
etary policy continues to target inflation, while transfers policy passively adjusts to stabilize
debt. Debt is stabilized when the government’s delivered transfers are only a fraction, λt, of
promised transfers, zt.11 Transition paths for this policy regime appear as triangles in figure
7. The figure shows paths from 2045 on and repeats the paths in figure 6, solid lines, for
comparison.
When the passive transfers policy is realized, the tax rate is below the fiscal limit, so
τt jumps immediately to τmax, creating a surprise increase in the tax rate. This causes a
sharp reduction in labor supply, hours worked, and output. Passive transfers implies that
the government does not fulfill its promises on transfer payments, as figure 8 shows. The
shock of the regime change drives actual transfers to 75 percent of promised and then the
fraction smoothly declines as the passive transfers regime continues in effect.
Regime change produces an unexpected drop in inflation, as agents had put a 50 percent
11Because transfers are lump-sum and λt is endogenous, there is a fundamental indeterminacy between thevalue of debt at the fiscal limit and the present value of λtzt. Given the process for zt, for each sequence ofλt ∈ [0, 1]∞
t=T, there is a corresponding value of debt at the limit, BT /PT . We resolve this indeterminacy
by assuming that the debt-output ratio settles in at 1.08 in the passive transfers regime.
31
“Unfunded Liabilities” and Uncertain Fiscal Financing
2010 2015 2020 2025 2030 2035 2040 2045 2050 2055 20600.5
1
1.5
2
2.5
3
PM/AF/AT
AM/AF/PT
AM/PF/AT
Figure 9: Ex-ante real interest rates in the three regimes. Active monetary/passivetax/active transfers policies are solid line; passive monetary/active tax/active transfers poli-cies are dashed lines; active monetary/active tax/passive transfers policies are dotted-dashedlines.
chance on going to the passive monetary regime in which inflation is higher. Active monetary
policy reacts to the drop in inflation by sharply lowering the nominal and the ex-ante real
interest rates [dotted dashed line in figure 9]. Despite the decline in output, initially lower
real rates raise consumption. As inflation gradually rises, the real interest rate rises, and
consumption smoothly declines. Gradually rising inflation stems from agents’ beliefs that
a switch to passive monetary policy—and the jump in inflation that it engenders—remains
a possibility. Even though the passive transfers regime is far more persistent (expected
duration of 100 years) than is the passive monetary regime (expected duration of 10 years),
as we see below, the jump in inflation that passive monetary policy produces is sufficiently
large that its impacts spill into the regime with active monetary and passive transfers policies.
Once again we see that the possibility that exploding promised transfers will be realized—
coupled with active monetary policy—prevents the central bank from effectively targeting
inflation.
5.1.3 Passive Monetary, Active Tax, and Active Transfers Policies The
final scenario has the government delivering on its promised transfers—active policy—while
monetary policy abandons its effort to target inflation, reverting instead to pegging the
nominal interest rate—passive monetary policy. In this regime, although promised transfers
32
“Unfunded Liabilities” and Uncertain Fiscal Financing
are currently being honored, agents place substantial probability on moving to the regime in
which the government will renege. In anticipation of that reneging, agents increase savings
dramatically, raising the capital stock well above even its initial 2009 level [marked by squares
in figure 7]. The initial jump in the ex-ante real interest rate creates the incentive to increase
investment [dashed line in figure 9].
At the time of the regime change there is a huge spike in inflation triggered by a sharp
downward revision in the expected present value of primary surpluses. Agents had put 50
percent probability on the passive transfers regime, but the realization of active transfers
causes them to revise downward their view of future surpluses. At the price level in place
before the regime change, the real value of outstanding nominal debt is now too high to be
supported by the new view of the surplus path. This creates a jump in aggregate demand—
debt is initially overvalued so households seek to reduce their debt holdings—which raises
labor demand, hours worked, output, and inflation. Costly price adjustment, however, more
than absorbs the additional output, so both investment and consumption fall in the year of
the regime change.12
Each period that policy remains in the passive monetary/active transfers regime, transfers
are unexpectedly high, relative to the expectation of switching to the more persistent active
monetary/passive transfers regime, another example of how expectations formation effects
flow across regimes. These surprises create a sequence of positive forecast errors in inflation
which steadily raise the inflation rate and, with monetary policy pegging the nominal interest
rate, reduces ex-post real interest rates. Lower debt service, coupled with the revaluation of
outstanding debt, helps to stabilize debt.
Based on the analytical results in section 4, it might seem that this combination of passive
monetary/active transfers policies, if it persisted indefinitely, could stabilize debt. That
outcome, though, relies on the real interest rate being exogenous. In the more sophisticated
model, capital accumulation drives down the real interest rate—see figure 9—which raises the
12Introduction of long-term nominal government bonds, along the lines of Cochrane (2001), would mitigatethese extreme one-time jumps.
33
“Unfunded Liabilities” and Uncertain Fiscal Financing
expected present value of the already explosive transfers process. A higher expected present
value of transfers, in turn, raises the expected rate of reneging in the active monetary/passive
transfers regime to which agents are expecting to transit in the future. But lower expected
future transfers encourages still more capital accumulation and the process repeats. The
presence of expectations formation effects, stemming from the prospect of the government
possibly reneging on promised transfers, makes the passive monetary/active transfers regime
unsustainable in the long run.
In this regime, as in the past two, monetary policy continues to lose control of inflation.
As figure 9 shows, the ex-ante real interest rate falls steadily after its initial jump at the time
of the regime change. At the same time, though, monetary policy is pegging the nominal
interest rate. Expected inflation must be rising. The combination of passive monetary policy
with an explosive promised—and honored—transfers process implies that monetary policy
not only cannot control actual inflation, but even the pegged nominalt interest rate fails to
control expected inflation.
5.2 Stationary Distribution of the Economy Figure 10 plots the 25th and 75th
percentiles bands (dashed lines) and the 10th and 90th percentiles bands (solid lines) of time
paths of variables in the stationary distribution of the economy. The economy was simulated
using 10,000 draws, assuming that year 2009 is the zero-shock steady state, stationary trans-
fer regime. The draws represent percentage deviations from the steady state distribution
over time. At each date, the figure depicts the cross-sectional distribution of the variables.
Several findings emerge.
First, the dispersion in the distribution and the deviation from 2009 levels is very small
for the first 10 years (from 2010–2020) for all variables. This result is not entirely inconsistent
with CBO projections [Congressional Budget Office (2009a)] and figure 2. According to the
CBO, the recovery of the economy and the corresponding increase in tax revenue, coupled
with the reduction in non-interest government spending to pre-financial crisis levels (see figure
1), stabilizes debt over the short term. However, the mechanism operating in our model is
34
“Unfunded Liabilities” and Uncertain Fiscal Financing
2010 2020 2030 2040 2050 2060
−6
−4
−2
0
Output (% deviation from 2009)
2010 2020 2030 2040 2050 2060
−10
0
10
Capital Stock (% deviation from 2009)
2010 2020 2030 2040 2050 20600.2
0.22
0.24
0.26
Tax Rate
2010 2020 2030 2040 2050 2060
0.5
1
1.5Debt / Output
2010 2020 2030 2040 2050 20601
1.5
2
2.5
3
3.5Inflation (%)
2010 2020 2030 2040 2050 20600.6
0.8
1
% of Promised Transfers Paid
Figure 10: Cross sections of stationary distribution of model variables from monte carlosimulation taking draws from policy regimes. Dashed lines are 25th and 75th percentilebands; solid lines are 10th and 90th percentile bands. Output, consumption, capital stock,and labor are percentage deviations from their 2009 levels. Based on 10,000 draws.
quite different. While there is always a chance of hitting the fiscal limit, the probability of
doing so is very small initially (2 percent) and increases only gradually. Agents know that
while the process for transfers is unsustainable and a policy adjustment must occur in the
future, the future is sufficiently far enough away that the expectational effects associated
with policy uncertainty are heavily discounted. This result is, of course, very sensitive to
the parameter values of the model. An increase in either the initial probability of hitting
the fiscal limit or a steepening of the logistic function so that this probability increases at
a faster rate over time, will change this result quantitatively. But qualitatively, the initial
impact of “unfunded liabilities” will be small relative to the effects several decades out.
Related to this point, figure 10 shows that the debt/output ratio never reaches 150
percent at any point in time, even for the 90th percentile band. This result is in stark
contrast to CBO projections in figure 2, where the debt-output ratio reaches a maximum
of well over 700 percent. As emphasized in the introduction, “unfunded liabilities” that
35
“Unfunded Liabilities” and Uncertain Fiscal Financing
are unsustainable are inconsistent with a rational expectations equilibrium. Assuming that
policies adjust to ensure a stationary equilibrium places an upper bound on the debt-output
ratio, but quantitatively the upper bound for our calibration only puts the debt-output
ratio slightly above that after World War II. Moreover, and quite importantly, this ratio is
achieved without a large amount of reneging on promised transfers. Figure 10 shows that
most of the probability mass for the promised transfers paid out is greater than 0.8 for the
entire time horizon, and there is no reneging at any percentile until 2030. As shown in figure
1, the calibrated distribution for non-stationary transfers roughly matches the projection of
the CBO, and assumes real transfers per capita grow at 1 percent per year. Even after the
fiscal limit is reached, our model allows for the possibility that promised transfers be paid in
full for some time, assuming that monetary policy becomes passive. This mechanism adds
an important dimension of reality to the model as policymakers may be reluctant to change
popular entitlement programs and avoid the political “third rail.” This mechanism explains
why the percentage of promised transfers paid is so high.
By leaving the door open for actual transfers to be less than promised, the model does not
produce the dire inflationary—or hyperinflationary—scenarios that some observers suggest
[Kotlikoff and Burns (2004)]. In figure 10, even the 90th percentile band never exceeds 3.5
percent annual inflation. These bands, though, do not report the one-time spikes that arise
when the passive monetary policy regime is realized.
Finally, the monte carlo exercise makes clear the time-varying uncertainty facing the
economic agents and the wide range of possible equilibrium outcomes. For example, figure
10 shows that the capital stock may rise or fall over time. An increase in the capital stock is
explained by a negative wealth effect operating through expected reneging on transfers. As
the probability of hitting the fiscal limit increases, there is an increase in the probability of
the reneging regime becoming reality. A precautionary savings result ensues as agents save
dramatically to offset the expected decline in transfers. The negative response of capital
that shows up as early as 2020 is due to the distorting effects of taxes. Draws of the tax
36
“Unfunded Liabilities” and Uncertain Fiscal Financing
rate in the 90th percentile spike to τmax prior to 2020 because the economy has hit the fiscal
limit. This “early” and persistent jump in the tax rate has the usual strong and negative
impact on capital formation and output.
6 Concluding Remarks
A key element of our analysis, and of any reasonable analysis, of these looming fiscal issues is
that future policy is marked by tremendous uncertainty. By taking a stand on the nature of
that uncertainty, this paper shows that it is possible to use a rational expectations equilibrium
to try to understand how these large fiscal issues will impact the macro economy. This is
the paper’s primary contribution.
To produce the numerical results we report, we have had to specify parameters describing
private and policy behavior. Results are highly sensitive to those parameter settings. The
framework the paper develops, however, is general and sufficiently flexible to accommodate
alternative specifications on private behavior and the potential range and nature of policy
adjustments in the future.
Despite the specificity of the stands we have taken, we are comfortable drawing some
broad conclusions:
1. Although the looming fiscal issues are large, they may not produce the “end-of-the-
world” scenarios that some commentators have suggested. For example, although in
this environment monetary policy can no longer achieve its inflation target, there is no
reason to expect a prolonged period of hyperinflation or even extremely high inflation.
While it is true that private consumption falls over the coming decades, GDP may
not be much lower, and the capital stock could even grow. In such an equilibrium,
the degree to which the government reneges on its promised transfers could be quite
moderate.
2. Uncertainty about future policies pushes many of the adjustments of the economy into
37
“Unfunded Liabilities” and Uncertain Fiscal Financing
the future, smoothing out adjustments that take place in the near term. Resolving
that uncertainty is likely to bring effects forward in time.
3. When government spending policies push tax policy to the fiscal limit—whether that
limit be economic or political—monetary policy loses control of inflation. Analyses of
monetary policy going forward must come to terms with this reality.
We have considered only a small set of possible policy scenarios. Many other scenarios
are possible. It is a worthwhile research program to examine other scenarios and trace out
their likely consequences for the macro economy.
This paper has performed no welfare analysis and does not examine distributional issues.
But the framework certainly points toward a research program that aims to design and
evaluate monetary and fiscal rules that can both cope with the coming fiscal issues and reduce
uncertainty about future policy. Of course, the political process—like research economists—
must ultimately take a stand on how to resolve America’s fiscal problems.
38
“Unfunded Liabilities” and Uncertain Fiscal Financing
A Appendix
A.1 Details About Figures 1 and 2 Figure 2 plots the actual and projected debt-to-
GDP ratio from 1790 through 2083. The dashed lines represent CBO projections [Congres-
sional Budget Office (2009b)] under two scenarios—the Alternative Fiscal Financing (AF)
and Extended-Baseline scenario (EB). The difference between the two forecasts is that the
EB projection assumes current law does not change, while the AF allows for “policy changes
that are widely expected to occur and that policymakers have regularly made in the past,”
[Congressional Budget Office (2009b)]. The prominent changes include: [i] assuming the ex-
piring tax provisions in the Economic Growth and Tax Relief Reconciliation Act of 2001 and
the Jobs and Growth Tax Relief Reconciliation Act of 2003 will be extended beyond 2010; [ii]
assuming the Alternative Minimum Tax will be indexed to inflation; [iii] assuming physician
payment rates under Medicare will grow at the same rate as the Medicare economic index.
A.2 Numerical Solution Method We solve the model using the monotone map
method from Coleman (1991). The algorithm is as follows:
1. Discretize the state space around the nonstochastic steady state for each state variable
(i.e. Θt = Kt−1, zt, bt−1, Rt−1, mt−1, Sz,t, St).
2. Conjecture an initial set of decision rules for the capital stock, labor, marginal costs
and inflation. We initially solved a simplified version of the model with stationary transfers
and either AM/PF or PM/AF policy. We then used weighted combinations of these rules as
the initial conjectures. Denote the initial rules as hKj (Θt) = Kt, hNj (Θt) = Nt, h
ψj (Θt) = ψt
and hπj (Θt) = πt for j = 0. We can then define decision rules for the other endogenous
variables using resource constraints.
3. At each point in the state space, substitute these decision rules into the household’s
first-order necessary conditions. In turn, the t + 1 dated endogenous variables depend on
Θt+1. Numerical integration is used then to integrate over the exogenous variables zt+1, Sz,t+1
39
“Unfunded Liabilities” and Uncertain Fiscal Financing
and St+1. For a given point on the state space, this procedure yields a nonlinear system of
equations. Solving this system for state variables at each point in the state space yields
updated values for the decision rules (e.g. hKj+1(Θt) = Kt). Note that the time t solution for
the system is used to construct the state vector for t+ 1.
4. Repeat step (3) until the decision rules converge at every point in the state space
(|hKj (Θt)− hKj+1(Θt)| < ǫ).
As evidence of local uniqueness, we perturb the converged decision rules in various di-
mensions and check that the algorithm converges back to the same solution. We restrict
our attention to solutions on the minimum set of state variables. The only channel for
states outside of the minimum set to matter is through expectation formation, so introduc-
ing additional states would require a solution method that allows latitude to specify how the
extraneous states affect expectations. Given there is no theory to guide such decisions in
the current setting, the discipline imposed by focusing on the minimum set of state variables
and the monotone map algorithm is appealing.
The monotone map computes a solution using only first-order necessary conditions, so
does not formally impose transversality conditions. In the present context, the transversality
condition plays a key role because it prevents the government from perputually rolling over
government debt. To gain insight into this issue, we solved simpler models using the mono-
tone map that explicitly violated the transversality condition. The algorithm converged in
these cases; however, the resulting simulations imply non-stationary paths for debt. Con-
versely, models that satisfied the transversality condition implied stable debt dynamics. As
a numerical check of the transversality condition, we simulated the model 5,000 times - ran-
domly drawing from policy regimes - and computed the average debt level at each date. We
found that after 46 periods the expected value of debt peaks at roughly 80 percent of output,
indicating that the expected value of debt does not asymptotically explode.13
13We cannot run similations out for, say, 10,000 periods as a check of the transversality condition becausethe promised transfer process is nonstationary. Given that promised transfers are a state variable, theapproximate solution will become increasingly inaccurate as the equilibrium dynamics drift further off thegrid.
40
“Unfunded Liabilities” and Uncertain Fiscal Financing
References
Abel, A. B., and O. Blanchard (1983): “An Intertemporal Equilibrium Model of Saving
and Investment,” Econometrica, 1(2), 675–692.
Aguiar, M., and G. Gopinath (2006): “Defaultable Debt, Interest Rates and the Current
Account,” Journal of International Economics, 69(1), 64–83.
Arellano, C. (2008): “Default Risk and Income Fluctuations in Emerging Economies,”
American Economic Review, 98(3), 690–712.
Auerbach, A. J., J. Gokhale, and L. J. Kotlikoff (1994): “Generational Accounting:
A Meaningful Way to Evaluate Fiscal Policy,” Journal of Economic Perspectives, 8(1), 73–
94.
(1995): “Restoring Generational Balance in U.S. Fiscal Policy: What Will It
Take?,” Federal Reserve Bank of Cleveland, 31(1), 3–12.
Auerbach, A. J., and K. Hassett (1992): “Tax Policy and Business Fixed Investment
in the United States,” Journal of Public Economics, 47, 141–170.
(2007): “Optimal Long-Run Fiscal Policy: Constraints, Preferences, and the Res-
olution of Uncertainty,” Journal of Economic Dynamics and Control, 31(5), 1451–1472.
Auerbach, A. J., and L. J. Kotilikoff (1987): Dynamic Fiscal Policy. Cambridge
University Press, Cambridge, M.A.
Bi, H. (2009): “Sovereign Risk Premia, Fiscal Limits and Fiscal Policy,” Manuscript, Indi-
ana University.
Bizer, D. S., and K. L. Judd (1989): “Taxation and Uncertainty,” American Economic
Review, 79(2), 331–336.
41
“Unfunded Liabilities” and Uncertain Fiscal Financing
Butler, M. (1999): “Anticipation Effects of Looming Public-Pension Reforms,” Carnegie-
Rochester Conference Series on Public Policy, 50, 119–159.
Chari, V. V., P. J. Kehoe, and E. R. McGrattan (2000): “Sticky Price Models
of the Business Cycle: Can the Contract Multiplier Solve the Persistence Problem?,”
Econometrica, 68(5), 1151–1179.
Cochrane, J. H. (1999): “A Frictionless View of U.S. Inflation,” in NBER Macroeconomics
Annual 1998, ed. by B. S. Bernanke, and J. J. Rotemberg, vol. 14, pp. 323–384. MIT Press,
Cambridge, MA.
(2001): “Long Term Debt and Optimal Policy in the Fiscal Theory of the Price
Level,” Econometrica, 69(1), 69–116.
Coleman, II, W. J. (1991): “Equilibrium in a Production Economy with an Income Tax,”
Econometrica, 59(4), 1091–1104.
Congressional Budget Office (2002): A 125-Year Picture of the Federal Government’s
Share of the Economy, 1950 to 2075. CBO, Washington, D.C.
(2009a): An Analysis of the President’s Budgetary Proposals for Fiscal Year 2010.
CBO, Washington, D.C., June.
(2009b): The Long-Term Budget Outlook. CBO, Washington, D.C., June.
Davig, T., and E. M. Leeper (2006): “Fluctuating Macro Policies and the Fiscal Theory,”
in NBER Macroeconomics Annual 2006, ed. by D. Acemoglu, K. Rogoff, andM. Woodford,
vol. 21, pp. 247–298. MIT Press, Cambridge.
(2007): “Generalizing the Taylor Principle,” American Economic Review, 97(3),
607–635.
Dotsey, M. (1990): “The Economic Effects of Production Taxes in a Stochastic Growth
Model,” American Economic Review, 80(5), 1168–1182.
42
“Unfunded Liabilities” and Uncertain Fiscal Financing
Hall, R. E. (1971): “The Dynamic Effects of Fiscal Policy in an Economy with Foresight,”
Review of Economic Studies, 38(2), 229–244.
Hassett, K., and G. Metcalf (1999): “Investment with Uncertain Tax Policy: Does
Random Tax Policy Discourage Investment,” The Economic Journal, 109(457), 372–393.
Imrohoroglu, A., S. Imrohoroglu, and D. H. Joines (1995): “A Life Cyle Analysis
of Social Security,” Economic Theory, 6(1), 83–114.
Kotlikoff, L. J., and S. Burns (2004): The Coming Generational Storm: What You
Need to Know About America’s Economic Future. MIT Press, Cambridge, MA.
Kotlikoff, L. J., K. Smetters, and J. Walliser (1998): “The Economic Impact of
Transiting to a Privatized Social Security System,” in Redesigning Social Security, ed. by
H. Siebert, p. 327. Kiel University Press, Kiel.
(2007): “Mitigating America’s Demographic Dilemma by Pre-Funding Social Se-
curity,” Journal of Monetary Economics, 54(2), 247–266.
Leeper, E. M. (1991): “Equilibria Under ‘Active’ and ‘Passive’ Monetary and Fiscal Poli-
cies,” Journal of Monetary Economics, 27(1), 129–147.
Mendoza, E. G., and V. Z. Yue (2008): “A Solution to the Default Risk–Business Cycle
Disconnect,” NBER Working Paper No. 13861.
North, D. C. (1990): Institutions, Institutional Change, and Economic Performance. Cam-
bridge University Press, Cambridge.
O’Neill, J. E. (1996): “CBO Testimony: The Long-Run Budgetary Impacts of an Aging
Population,” Statement before the Committee on the Budget, U.S. House of Representa-
tives, Congressional Budget Office, March 13.
Peter G. Peterson Foundation (2009): State of the Union’s Finances: A Citizen’s
Guide. www.pgpf.org, March.
43
“Unfunded Liabilities” and Uncertain Fiscal Financing
Rotemberg, J. J. (1982): “Sticky Prices in the United States,” Journal of Political Econ-
omy, 90(December), 1187–1211.
Rotemberg, J. J., and M. Woodford (1997): “An Optimization-Based Econometric
Framework for the Evaluation of Monetary Policy,” in NBER Macroeconomics Annual
1997, ed. by B. S. Bernanke, and J. J. Rotemberg, vol. 12, pp. 297–346. MIT Press,
Cambridge, MA.
Sanger, D. E. (2010): “Long-Term Debt. Obama Budget Sees Years of Deficits, A Test of
U.S. Power and Political Will,” New York Times, February 2, 2010, p. A1.
Sargent, T. J. (2006): “Ambiguity in American Monetary and Fiscal Policy,” Japan &
The World Economy, 18(3), 324–330.
Sargent, T. J., and N. Wallace (1981): “Some Unpleasant Monetarist Arithmetic,”
Federal Reserve Bank of Minneapolis Quarterly Review, 5(Fall), 1–17.
Sims, C. A. (1994): “A Simple Model for Study of the Determination of the Price Level
and the Interaction of Monetary and Fiscal Policy,” Economic Theory, 4(3), 381–399.
Smetters, K., and J. Walliser (2004): “Opting Out of Social Security,” Journal of
Public Economics, 88(7-8), 1295–1306.
Woodford, M. (1995): “Price-Level Determinacy Without Control of a Monetary Ag-
gregate,” in Carneige-Rochester Conference Series on Public Policy, vol. 43 of Carnegie-
Rochester Conference Series on Public Policy, pp. 1–46. North-Holland.
(2003): Interest and Prices: Foundations of a Theory of Monetary Policy. Princeton
University Press, Princeton, N.J.
44